fhein

Lite fredagsunderhållning åt folket

Couldn't find a dedicated community for distro recommendations, I hope it's ok to ask here.

A couple of years ago my wife and I built a computer and gave it to a friend's kid. We put ElementaryOS on it since that seemed pretty fool-proof, but it appears to require a re-install to upgrade major versions so it has been stuck with an old glibc and because of that he can't play Factorio.

For his 13:th birthday we bought him a SSD so it would be a good time to reinstall Linux, but is there perhaps some better choice than ElementaryOS? They live quite far away so I can't easily pop over to fix his computer if something breaks, we don't spend enough time there for me to teach him to fix things himself, and he doesn't seem very interested in learning how computers/operatings systems work either.

• Hardware: Some old Intel CPU with 8GB DDR3 and a GTX1080
• Usage: Gaming through Steam+Proton, Lutris and browsing.
• Requirements: Games work, OS never breaks on updates. Doesn't need to be "kid proof", I don't think he touches any stuff he doesn't know what it does.

In case anyone isn't familiar with llama.cpp and GGUF, basically it allows you to load part of the model to regular RAM if you can't fit all of it in VRAM, and then it splits the inference work between CPU and GPU. It is of course significantly slower than running a model entirely on GPU, but depending on your use case it might be acceptable if you want to run larger models locally.

However, since you can no longer use the "pick the largest quantization that fits in memory" logic, there are more choices to make when choosing which file to download. For example I have 24GB VRAM, so if I want to run a 70B model I could either use a Q4_K_S quant and perhaps fit 40/80 layers in VRAM, or a Q3_K_S quant and maybe fit 60 layers instead, but how will it affect speed and text quality? Then there are of course IQ quants, which are supposedly higher quality than a similar size Q quant, but possibly a little slower.

In addition to the quantization choice, there are additional flags which affect memory usage. For example I can opt to not offload the KQV cache, which would slow down inference, but perhaps it's a net gain if I can offload more model layers instead? And I can save some RAM/VRAM by using a quantized cache, probably with some quality loss, but I could use the savings to load a larger quant and perhaps that would offset it.

Was just wondering if someone has already done experiments/benchmarks in this area, did not find any exact comparisons on search engines. Planning to do some benchmarks myself but not sure when I have time.

Update: Bug fixed in Plasma 6.3.1

Just posting this since I spent over an hour trying to figure out why I couldn't open my desktop today.. After booting and logging in I got a black screen. Switched to a terminal but did not see any obvious errors in the logs.

Not fixed for Fedora 41 KDE yet, so I installed plasma-workspace-x11 to use in the meanwhile. Anyone who hasn't updated to 6.3 yet could probably change their display settings to not use ICC profiles to avoid it.___

I just spent half an hour trying to figure this out so I thought I'd write it down somewhere in case it helps someone else in the future.

Aslain's modpack contains a whole lot of quality-of-life mods for WoWs, for example Battle Expert (formerly known as Navigator) which shows the exact relative angles between your ship and the enemy's. Almost feels like cheating to me, but Wargaming has endorsed this modpack and it even has a dedicated channel on the official discord server. Theoretically you have the same information without the mod, but it can be difficult to see how a ship is turning or changing speed by just looking at it.

These instructions are for when the game is installed through Steam, which looks like it uses some kind of overlay filesystem. This led to that the game install folder didn't show up for the modpack installer when I tried other methods.

1. Install protontricks, I used the version available in Fedora's repos.
2. Download the modpack installer from the official site
3. Find the WoWs install folder in Steam. Right-click World of Warships in the Steam games list, select Manage and "Browse local files" and the folder should open in your default file manager.
4. In a terminal, run the modpack installer .exe file in the game's Wine prefix. I'm not entirely sure this makes any difference compared to running it in a new prefix as long as it can access the game files, it mostly seemed convenient to me. The app id for WoWs is 552990 and it should never change, but you can get it with protontricks -l if you're curious. Change the file path so that it matches the file you downloaded and run:
   protontricks-launch --appid 552990 ~/Downloads/Aslains_WoWs_Modpack_Installer_v.13.6.1_01.exe
   It will print a lot of "failed to create" error messages for system dlls and exes, but that appears to be normal, and the setup window should open after a while.
5. After some release notes etc. the installer will eventually ask you for the game's install dir. As far as I can tell, the game files do not show up anywhere on C:, but Steam mounts your Linux file system on Z: so we can use that instead. Browse to the game install folder, which we located in step 3, and select it. My install folder on Linux is
   /mnt/faststore/SteamLibrary/steamapps/common/World of Warships/ so I select
   Z:\mnt\faststore\SteamLibrary\steamapps\common\World of Warships in the modpack installer.
6. Either manually select the mods you want or use the recommended selection. As I wrote before, many for these mods feel like they give you an in-game advantage over other players, but WG has said they're legal...
7. The first time I ran the installer it hung on "Finishing installation". It appears to happen to a few Windows users too but the mod dev doesn't know what causes it. I noticed that there was a cleanup process running in Wine C:\windows\system32\cmd.exe /C DEL /s /f *.orig which shouldn't take so long time so I killed it (in Linux) and the installer continued. The next time I ran it this didn't happen, and it only took a few seconds to finish the installation.

If you have the game installed as standalone, e.g. Lutris, then I think you can just run the modpack installer in the same Wine prefix, and you should see the game's install folder under C:\Program Files as you would on Windows. I.e. select the game in Lutris, click the tiny arrow next to the wine glass button and select "Run EXE inside Wine prefix" and then choose the installer you downloaded. But I haven't done this so I promise nothing.

Please don't take this as an endorsement of World of Warships, I borderline hate this game and only play it because some of my friends are obsessed with it. The gameplay is a bit too slow paced for my taste, there are a lot of hard counters which you can't do anything about in random matchmaking, and carriers (planes) can turn any game into pure suffering. I also dislike the game's monetization scheme, lootboxes are expensive and most have a tiny chance to give something really good and a big chance to give you complete garbage. The game might be f2p, but at higher tiers it becomes unplayable without a premium subscription (€10/month) since ship maintenance gets more expensive than your earnings. To maximize your ship's performance you need a high level captain, expensive modules and also buffs which are consumed each game. My friend tries to argue that the game is not pay-to-win because you can also grind ingame resources to buy those, but you'll spend many hours playing at a disadvantage if you don't buy your way past it. Just my personal opinion of course.

If you despite my warnings felt an urge to try this game (honestly I thought it was quite fun at lower tiers) then check if any of your friends are already playing it and ask them for a referral code. Both of you get free stuff from being recruited by someone else and once you've created an account it's too late, unless you stop playing completely for 3 months. If you do that it is possible for your friend to send you a recruiting link if you want to start playing again.

Just a heads up, I've read that it's impossible to connect an existing wargaming.net account to a Steam account on Linux, so make sure you authenticate through Steam when you create the account if you plan on playing it through Steam. Though if you have Windows dual boot then I think you can link the accounts there if you need to.

Going through some boxes and found a stack of old White Dwarf. I'll keep the first issue I ever bought as a memory but planning to get rid of the rest. Just wanted to check if there are people collecting these before they go into the recycling bin. If anyone's interested I can make a list of which ones I have, and I'll send them to anyone willing to pay for postage. Located in Sweden.

Any games with less than 1000 total Steam reviews you've enjoyed and thought more people ought to know about? Not a hard limit, just a guideline for what could be classified as "undiscovered" on Steam, assuming it wasn't released yesterday.

I would recommend:

• Full Bore, a cute block-based puzzle platformer. Solid mechanics, level designs and even a somewhat engaging story. ~~Unfortunately hasn't been on a sale since 2021 according to steampricehistory.com, while it was frequently reduced to €2-3 before that. Not sure I'd recommend it to everybody at full price, but IMO it's one of the best indie platformers I've played.~~ edit: Did someone email the creator of Full Bore or something? It's suddenly on sale again, for the first time in ages :) Go buy it!

cross-posted from: https://lemmy.world/post/1396665

I'm about to invest in some "prosumer" level 18V power tools, e.g. Makita, Dewalt, Metabo, etc. The general consensus appears to be that they're all more or less equal and it doesn't matter much which brand one goes with, but a few years ago I developed tinnitus which has made me more sensitive to high frequency sounds. And having tested a few cordless drill drivers at various hardware stores I can definitely tell they're not equal when it comes to what the noise sounds like, even if they measure similarly in sound pressure level.

Currently I have a Ryobi 18V brushless that I measure to 83-84 dB(c) at 1m which makes it pretty loud, and it also has a quite a quite high pitched sound which makes it worse. Looking at a spectrogram from a calibrated measurement mic there's a pretty clear peak at 6kHz. The drill drivers I've been looking at all have a rated sound pressure level (Lp) around 76 dB so they should already be noticeably less loud, but I'm hoping to find one without those high frequencies if possible.

I've been able to test a few drills in person and got some initial impressions. Unfortunately the local hardware stores have a pretty limited selection of brands, they don't have all models in the store, and you usually have to bother the staff if you want to try something with batteries in it. At first I thought I could remember my subjective opinion of different drivers, but when I later got the opportunity to re-compare two of them head to head I realised I have really bad memory..

Head-to-head comparisons that I've done:

Makita DDF484 vs Dewalt DCD791: The Makita has a very high frequency whine, kind of like dentist's drill. The Dewalt is probably just as loud, but a little lower pitch making the noise slightly more tolerable, while still not great.

Makita DDF484 vs 485 vs 486: When it comes to high pitched noise, 484 was worst, followed by 485, and the 486 was most tolerable.

Dewalt DCD791 vs DC800: The 800 had a more high pitched noise than the 791, once again reminding me of the dentist's drill.

Not compared with anything, so only subjective impression:

Hikoki (a.k.a. Metabo HPT in USA) DV18DD and Dewalt DCD777: Smaller and weaker tools but much more quiet than everything else. If I were going to buy 2 drill drivers I'd probably get something like these plus one larger for heavier work.

Metabo (the made in Germany, non-HPT, brand) DS 18 LT BL: Only had a quick try with this machine but without having anything to compare it with I thought it sounded pretty reasonably tolerable. There's a version with quick change chucks, unfortunately only the smaller (L) and larger (LTX) models have offset and angle chucks for some reason but that's a feature I would really like to have.

Anyone compared Milwaukee to other brands with respect to noise? The only store I found that sells them said they didn't have any charged batteries so I couldn't try any.

Not exactly sure what I'm expecting from this post, since the question is so subjective.. Perhaps someone with experience of multiple cordless drill drivers could share their opinion? Or someone else with a similar aversion to high pitched noises could say if they found any tool that they're happy with?

I've been toying with the idea of replacing all my old, mostly corded, power tools with a single 18V system and like many others I'm trying to decide between Makita and Dewalt. They seem to be pretty even and most recommendations boil down to "pick the colour you like best".

However I bought a subscription to a magazine that does a lot of product testing, and saw that they gave all Makita drills the lowest rating in the "expected battery life time" category because they only lasted 300 charging cycles out of the 400 they do as part of the test. Now 300 cycles is quite a lot for a home DIY:er, but is battery life something that Makita owners have had issues with? I can't remember seeing anyone complaining about it when looking for user reviews and comments. However I do have a vague memory of some pros complaining about Makita "smart" tools shutting down supposedly because they detect that something might break, thus forcing the owner to hand it in for service.

Article in Swedish, and probably paywalled.

edit: Since I doing a bit of testing how annoying (i.e. mostly how high pitched) different drill drivers sound, which hopefully could be useful to someone else, I've updated the title to include this.